Less Butter, More Plant Oils?

The latest from our friends in nutritional epidemiology. Should we be grateful? Part 2 of a rules of evidence controversy.

People who consume plant-based oil instead of butter may experience beneficial health effects and even have a lower risk of premature death...

"Even cutting back butter a little and incorporating more plant-based oils into your daily diet can have meaningful long-term health benefits," [Dr. Daniel] Wang said.

Massachusetts General Hospital press release, March 6, 2025

This month’s adventure in nutrition news must have the seed oil purveyors dancing in the aisles. It’s exactly the public relations boost that the industry needed, what with R.F. Kennedy Jr. driving health policy in this country and some newly influential physicians disseminating widely the notion that seed oils are toxic.

As for the study itself: “The researchers,” according to the press release, “examined diet and health data from 200,000 people followed for more than 30 years and found that higher intake of plant-based oils, especially soybean, canola, and olive oil, was associated with lower total, cancer, and cardiovascular disease mortality, whereas butter intake was associated with increased risk of total and cancer mortality.”

The authors of the study are epidemiologists whom we’ve come to think of as the usual suspects (h/t Casablanca) in this world—Walter Willett and his collaborators at the Harvard T.H. Chan School of Public Health (HSPH). Willett and the HSPH epidemiologists oversee multiple huge cohort studies, each comprising tens of thousands of medical professionals. And these studies have been generating associations for decades—since 1990, at least—suggesting that we would all be healthier if the fats and proteins we consumed came from plants rather than animals.

Along the way, Willett and his colleagues have become the most highly-cited and influential nutritionists in the world, and its their research that has driven the 30-year transformation of official dietary guidance from merely pro-vegetable to actively anti-meat (as I discussed in a post last December in the context of the USDA Dietary Guidelines).

For those who want the details on the latest butter-plant oil study, I recommend Vinay Prasad’s recent post, which pulled (per Prasad’s style) no punches. After deconstructing the technical details of the paper, he suggests that the HSPH interpretation of the evidence constitutes “research misconduct,” and adds, “This paper is not science; it is closer to propaganda.”

It is also, regrettably, par for the course in nutritional epidemiology. Indeed, in some ways it’s propaganda for nutritional epidemiology. Publication of the article on such a timely issue and in a respected journal is an implicit assertion that this observational research can tell us something meaningful about the diet-disease relationships; that it should be taken seriously.

Simply put, why bother publishing the paper other than to suggest we don’t need expensive clinical trials to know that seed oils are benign and butter is not? What’s to gain? What should we take away from this observational research in the absence of such trials?

Because these questions are critical to understanding not just nutrition science but also mainstream dietary beliefs and guidelines, I’m going to leave the analysis of this paper to Prasad, and take the opportunity instead to discuss the larger issues. How did we get into a situation in which our nutritional policy and guidance depends almost entirely on a type of research that the medical community and even experts in the methodology of the research itself consider unreliable?

This post continues a series I started a year ago with an article I wrote for Nina Teicholz’s Unsettled Science newsletter. For those who never read it, I’ve also posted it here on Uncertainty Principles.¹

Observational epidemiology vs. evidence based medicine

That original post in February 2024—“Less Meat, More Plants”—sets the stage for this discussion. It describes what happened in 2019 when an influential collaboration of evidence-based medicine (EBM) researchers used a widely-accepted tool known as GRADE (The Grading of Recommendations Assessment, Development and Evaluation) to assess the evidence for the more-plants-less-meat nutritional consensus.

The result was a series of articles published in Annals of Internal Medicine, concluding that the more-plants-less-meat guidance was supported only by the associations from the kind of observational studies carried out so prominently at the Harvard School of Public Health. The EBM researchers then dismissed this evidence as “low to very low quality,” as the GRADE methodology does.

They then concluded that the authorities counseling us to avoid meat and eat more plants are doing so based on faith not science. The evidence itself could support no recommendation other than “continue current… consumption.”

Unsurprisingly, the nutritionists and the nutritional epidemiologists vehemently objected. After all, the nutritionists had embraced and disseminated the mostly-plants dogma; the nutritional epidemiologists were the ones who had made their careers generating the allegedly low-to-very-low quality evidence.